{"id":30614530,"url":"https://github.com/aisingapore/bmad-aisg-aiml","last_synced_at":"2025-08-30T07:14:44.779Z","repository":{"id":309907175,"uuid":"1037451533","full_name":"aisingapore/bmad-aisg-aiml","owner":"aisingapore","description":"BMAD AI/ML Engineering Expansion Pack - Streamlined framework for AI Singapore programs (MVP, POC, SIP, LADP) with specialized agents, workflows, and templates for ML/LLM development","archived":false,"fork":false,"pushed_at":"2025-08-14T12:24:49.000Z","size":175,"stargazers_count":1,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2025-08-14T14:21:05.470Z","etag":null,"topics":["ai-ethics","ai-singapore","aiml","bmad","bmad-method","data-science","llm","machine-learning","ml-engineering","mlops","rag"],"latest_commit_sha":null,"homepage":"","language":null,"has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/aisingapore.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":"CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":"CODEOWNERS","security":null,"support":null,"governance":null,"roadmap":null,"authors":"AUTHORS","dei":null,"publiccode":null,"codemeta":null,"zenodo":null}},"created_at":"2025-08-13T15:37:01.000Z","updated_at":"2025-08-14T12:24:53.000Z","dependencies_parsed_at":"2025-08-14T14:21:11.513Z","dependency_job_id":"1286c72c-a1ed-4fc1-898a-3d44c4a95494","html_url":"https://github.com/aisingapore/bmad-aisg-aiml","commit_stats":null,"previous_names":["aisingapore/bmad-aisg-aiml"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/aisingapore/bmad-aisg-aiml","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aisingapore%2Fbmad-aisg-aiml","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aisingapore%2Fbmad-aisg-aiml/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aisingapore%2Fbmad-aisg-aiml/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aisingapore%2Fbmad-aisg-aiml/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/aisingapore","download_url":"https://codeload.github.com/aisingapore/bmad-aisg-aiml/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/aisingapore%2Fbmad-aisg-aiml/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":272817671,"owners_count":24998028,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","status":"online","status_checked_at":"2025-08-30T02:00:09.474Z","response_time":77,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["ai-ethics","ai-singapore","aiml","bmad","bmad-method","data-science","llm","machine-learning","ml-engineering","mlops","rag"],"created_at":"2025-08-30T07:14:43.312Z","updated_at":"2025-08-30T07:14:44.070Z","avatar_url":"https://github.com/aisingapore.png","language":null,"funding_links":[],"categories":[],"sub_categories":[],"readme":"# BMAD AI/ML Engineering Expansion Pack (v2.0)\n\nThis expansion pack extends the BMAD Method framework to support AI/ML engineering projects. It provides agents, workflows, templates, and best practices in a consolidated architecture.\n\n## Project History\n\n**Founder**: Laurence Liew ([@beowulf68](https://github.com/beowulf68)) - Developed the initial workflows and agent framework. His contributions established the core methodology and approach.\n\n**Current Maintainers**: \n- **Najib Ninaba** ([@najibninaba](https://github.com/najibninaba)) - Core Team Member\n- **Siavash Sakhavi** ([@ssakhavi](https://github.com/ssakhavi)) - Core Team Member\n\n## Project Timelines\n\nBMAD frameworks support different project timelines:\n- **Traditional Timeline**: 6-12 months\n- **BMAD Timeline**: 3-12 weeks\n- **Iteration Speed**: Prototyping and testing cycles\n- **Deployment Frequency**: Multiple deployments per day\n\n## Overview\n\nThe streamlined AI/ML Engineering expansion pack provides specialized agents, workflows, templates, and best practices for:\n\n- Machine Learning model development and deployment\n- Large Language Model (LLM) and RAG application development\n- Comprehensive MLOps pipeline implementation\n- Unified AI security, ethics, and governance\n- Data science and analytics workflows\n- AI Singapore program-specific workflows\n\n## Installation\n\n### Prerequisites\n\nYour target project must be BMAD-enabled with instructions from [BMAD-METHOD](https://github.com/bmad-code-org/BMAD-METHOD).\n\n### 1. Clone This Repository\n\n```bash\n# Clone the expansion pack\ngit clone https://github.com/aisingapore/bmad-aisg-aiml.git\ncd bmad-aisg-aiml\n```\n\n### 2. Install the [BMAD Pack Installer](https://github.com/najibninaba/bmad-pack-installer) via `uv`\n\n```bash\n# Install once (recommended for regular use)\nuv tool install bmad-pack-installer\n\n# Or run directly without installation\nuvx --from bmad-pack-installer bmad-pack-installer deploy . /path/to/project\n```\n\n### 3. Deploy This Pack\n\n```bash\n# Basic installation (from cloned repo directory)\nbmad-pack-installer deploy . /path/to/project\n\n# Preview changes without installing\nbmad-pack-installer deploy . /path/to/project --dry-run\n\n# Force reinstall over existing pack\nbmad-pack-installer deploy . /path/to/project --force\n```\n\n### Validation\n\n```bash\n# Check if target is valid BMAD project\nbmad-pack-installer check /path/to/project\n\n# Validate this expansion pack (from cloned repo directory)\nbmad-pack-installer validate .\n```\n\nThe installer creates:\n- Hidden directory: `.bmad-aisg-aiml/`\n- Claude commands: `.claude/commands/bmadAISG/`\n- Updated manifests and symbolic links\n\n## Usage\n\n### Follow the Workflow\n\n1. **Check workflows folder** for workflow files:\n   ```bash\n   # Navigate to installed workflows\n   cd .bmad-aisg-aiml/workflows/\n   ls\n   ```\n\n2. **For Claude Code implementation**: Run the agents and tasks manually using `/{agent-name}` command.\n\n## Core Agent Team\n\n### 5 Agents\n\n1. **Marcus Tan Wei Ming** - ML/AI Engineer \u0026 MLOps Specialist (`ml-engineer`)\n   - **Heritage**: Singaporean Chinese\n   - **Expertise**: End-to-end ML development, MLOps pipelines, infrastructure automation\n   - **Technical Skills**: PyTorch/TensorFlow, Kubernetes/Docker, CI/CD, cloud platforms\n   - **Focus Areas**: Model training, deployment, monitoring, production systems\n\n2. **Rizwan bin Abdullah** - ML/AI System Architect (`ml-architect`)\n   - **Heritage**: Singaporean Malay\n   - **Expertise**: ML system design, scalable architectures, infrastructure planning\n   - **Technical Skills**: Distributed systems, transformer architectures, RAG systems\n   - **Focus Areas**: System design, model architecture selection, technical strategy\n\n3. **Sophia D'Cruz** - Senior Data Scientist (`ml-data-scientist`)\n   - **Heritage**: Singaporean Eurasian\n   - **Expertise**: Statistical analysis, experimental design, recommendation systems\n   - **Technical Skills**: Causal inference, A/B testing, feature engineering\n   - **Focus Areas**: EDA, hypothesis testing, insights generation, model evaluation\n\n4. **Priya Sharma** - ML Security \u0026 Ethics Specialist (`ml-security-ethics-specialist`)\n   - **Heritage**: Singaporean Indian\n   - **Expertise**: ML security, adversarial testing, AI ethics, compliance\n   - **Technical Skills**: Red teaming, bias detection, privacy protection\n   - **Focus Areas**: Security audits, ethical reviews, regulatory compliance\n\n5. **Dr. Dylan Poh** - ML Research Scientist \u0026 Experimental Design Specialist (`ml-researcher`)\n   - **Heritage**: Singaporean Chinese\n   - **Expertise**: ML research planning, experimental design, literature review, hypothesis formulation\n   - **Technical Skills**: Advanced mathematics, ML frameworks, distributed computing, scientific writing\n   - **Focus Areas**: Research methodology, state-of-the-art ML techniques, reproducible experiments\n\n## Creating New Agents\n\n### 1. Use previous agents as a template\nUse previous agents as a template to create new agent folder:\n\nRun in Claude Code (example):\n```bash\nUse agents/ml-architect.md as a template to create a ml-researcher agent\n```\n\n### 2. Define Agent Commands in Prompt\nInsert commands under \"commands\" in agent file:\n\n```yaml\ncommands:\n  - help: Show numbered list of the following commands to allow selection\n  - literature-review: use task create-research-doc.md with literature-review-tmpl.yaml\n```\n\n#### Explanation for literature-review command\n- You should use formats like `use {task} with {template}` or `execute {task}` for the commands. (Refer to original bmad agents)\n- `create-research-doc.md` should be placed in the **tasks** folder\n- `literature-review-tmpl.yaml` should be placed in the **templates** folder\n\n#### Tips for creating commands\n- Original BMAD has generic tasks like `create-doc` and `advanced-elicitation` which are included in this package.\n- For complex tasks, generate your own task file\n\n## How It Works\n\n- The installer copies some folders into `.claude/commands` folder these are the files which `/{command}` run in bmad.\n- `/{agent-name}` Run the prompt from the `.claude/commands/bmad-expansion-name/agents/agent-name` file.\n\n- To enable the agent to locate your file ensure the file path is inside the agent prompt. (This should be done by the installer)\n- This is a protion of the original prompt from a BMAD agent.\n```yaml\nIDE-FILE-RESOLUTION:\n  - FOR LATER USE ONLY - NOT FOR ACTIVATION, when executing commands that reference dependencies\n  - Dependencies map to {root}/{type}/{name}\n  - type=folder (tasks|templates|checklists|data|utils|etc...), name=file-name\n  - Example: create-doc.md → .bmad-core/tasks/create-doc.md.   #This line tells the agent where the hidden folder to check\n  - IMPORTANT: Only load these files when user requests specific command execution\n```\n\n## Workflows\n\n### Standard ML Workflows\n- **ML Development**: End-to-end model development process\n- **ML Deployment**: Production deployment with monitoring\n- **ML Experimentation**: Systematic experimentation framework\n\n### AI Singapore (AISG) Program Workflows\n\n| Program | Duration | Team Structure | Deliverable | Key Difference |\n|---------|----------|---------------|-------------|----------------|\n| **MVP** | 6 months | 1 AI Engineer + 2-6 Apprentices | Full production system | Comprehensive with training |\n| **POC** | 3 months | 1 AI Engineer + 2-4 Apprentices | Proof of concept | Feasibility with training |\n| **SIP** | 3 months | 1-2 AI Engineers (NO apprentices) | Production MVP | Fast delivery, no training |\n| **LADP** | 4 months | Learners + Mentors (guide only) | LLM application | Self-directed learning |\n\n#### 1. 6-Month MVP Projects (`aisg-mvp-workflow`)\n- **Team**: 1 AI Engineer + 2-6 Apprentices\n- **Objective**: Build comprehensive production system with apprentice training\n- **Phases**: Discovery → Experimentation → Productionization → Validation\n- **All 4 agents** activated across phases\n\n#### 2. 3-Month POC Projects (`aisg-poc-workflow`)\n- **Team**: 1 AI Engineer + 2-4 Apprentices\n- **Objective**: Validate technical feasibility and business value\n- **Phases**: Rapid Discovery → Prototyping → Deployment → Validation\n- **All 4 agents** for comprehensive validation\n\n#### 3. 3-Month SIP - Short Industry Projects (`aisg-sip-workflow`)\n- **Team**: 1-2 AI Engineers only (NO apprentices)\n- **Objective**: Deliver production MVP without training overhead\n- **Phases**: Discovery → Development → Productionization → Handover\n- **All 4 agents** for fast MVP delivery\n\n#### 4. 4-Month LADP - LLM Application Developer Programme (`aisg-ladp-workflow`)\n- **Duration**: 4 months part-time (8-10 hrs/week) or 1-3 days full-time\n- **Team**: Learners with mentor guidance (mentors guide but DON'T code)\n- **Objective**: Build real-world LLM applications with company SOW\n- **Structure**: Month 1 (Self-learning) → Month 2 (Design) → Month 3 (Development) → Month 4 (Deployment)\n- **3 workshops** + project implementation\n\n### Program Outcomes\n- **MVP**: Production systems completed in 6 months with training\n- **POC**: Proof of concepts completed in 3 months with learning\n- **SIP**: Production MVPs completed in 3 months\n- **LADP**: LLM applications developed in 4 months\n\n### 100E User Story Generation Workflow\n\n```mermaid\n    graph TD\n        A[Start: AI/ML Project] --\u003e B[ml-architect: aiml-brief.md]\n        B --\u003e C[ml-researcher: literature-review.md]\n        B --\u003e D[ml-architect: aiml-design-document.md]\n        C --\u003e D\n        D --\u003e E[ml-architect: aiml-architecture.md]\n        E --\u003e F[ml-architect: user-stories.md]\n        F --\u003e G[ml-architect: shard documents]\n        G --\u003e H[ml-engineer: create story]\n        H --\u003e I[ml-engineer: validate story]\n        I --\u003e|Yes| H\n        I --\u003e|No| J[user: provide feedback]\n        J --\u003e H\n        \n        %% Styling with black font and unique colors for each agent\n        style A fill:#E8F5E8,color:#000000,stroke:#000000\n        style B fill:#FFE4E1,color:#000000,stroke:#000000\n        style C fill:#E6F3FF,color:#000000,stroke:#000000\n        style D fill:#FFE4E1,color:#000000,stroke:#000000\n        style E fill:#FFE4E1,color:#000000,stroke:#000000\n        style F fill:#FFE4E1,color:#000000,stroke:#000000\n        style G fill:#FFE4E1,color:#000000,stroke:#000000\n        style H fill:#FFF2CC,color:#000000,stroke:#000000\n        style I fill:#FFF2CC,color:#000000,stroke:#000000\n        style J fill:#F0E68C,color:#000000,stroke:#000000\n        \n        %% Color legend for agents:\n        %% ml-architect (Rizwan): #FFE4E1 (Light Coral)\n        %% ml-engineer (Marcus): #FFF2CC (Light Yellow)\n        %% ml-data-scientist (Sophia): #E6F3FF (Light Blue)\n        %% ml-security-ethics-specialist (Priya): #E8F8E8 (Light Green)\n```\n\n\n## 🇸🇬 Singapore Context\n\nAll agents include:\n- **Local regulatory knowledge**: PDPA, IMDA, MAS\n- **AISG program experience**: MVP, POC, SIP, LADP workflows\n- **Understanding of local market dynamics**: Singapore tech ecosystem\n- **Government standards compliance**: National AI governance standards\n\n## File Structure\n\n```\nbmad-ai-ml-engineering/\n├── agents/                    # 5 core agents\n│   ├── ml-engineer.md\n│   ├── ml-architect.md\n│   ├── ml-data-scientist.md\n│   ├── ml-security-ethics-specialist.md\n│   └── ml-researcher.md\n├── agent-teams/              # 5 team configurations\n├── checklists/              # 4 checklists\n├── templates/               # 8 templates\n├── tasks/                   # 5 tasks\n├── workflows/               # Standard + 4 AISG workflows\n├── data/                    # 2 reference files\n├── web-bundles/            # 5 ready-to-use bundles\n└── config.yaml             # Configuration\n```\n\n## 📋 Dependencies\n\n- **✅ Required**: bmad-core \u003e= 4.0.0\n- **🔧 Recommended**: Python \u003e= 3.8, Docker, Kubernetes\n- **➕ Optional**: Terraform, MLflow, Kubeflow\n\n## ⚖️ Compliance \u0026 Standards\n\n### Singapore Regulations\n- **PDPA**: Personal Data Protection Act compliance\n- **IMDA**: Model AI Governance Framework aligned\n- **MAS FEAT**: Fairness, Ethics, Accountability, Transparency\n\n### International Standards\n- ISO/IEC 23053: Framework for AI using ML\n- ISO/IEC 23894: AI risk management\n\n## 🤝 Contributing\n\nContribution process:\n- **Core Team**: Direct commit access for maintenance and development\n- **External Contributors**: Submit contributions via pull requests\n- **Review Process**: All PRs require approval from core team members\n\nSee our [Contributing Guidelines](CONTRIBUTING.md) for detailed information on how to contribute.\n\nFor a complete list of contributors, see [CONTRIBUTORS.md](CONTRIBUTORS.md).\n\n## 🎓 Training \u0026 Support\n\n### 📚 Documentation\n- **Quick Start**: This README\n- **Workflows**: `/workflows/README.md`\n- **Web Bundles**: `/web-bundles/WEB-BUNDLE-INSTRUCTIONS.md`\n- **Agents**: Individual agent files in `/agents/`\n\n### 🛠️ Support Channels\n- Review `REFACTORING-SUMMARY.md` for v2.0 changes\n- Check agent-specific documentation\n- Consult workflow guides\n- Raise issues in the repository\n\n---\n\n## 📝 Version History\n\n### v2.0.0 (Current)\n- **5 core agents** (added ml-researcher)\n- Added SIP workflow for MVP delivery\n- Updated LADP to 4-month programme\n- Added Singapore context\n\n### v1.0.0\n- Initial release\n- Basic AISG workflows\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Faisingapore%2Fbmad-aisg-aiml","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Faisingapore%2Fbmad-aisg-aiml","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Faisingapore%2Fbmad-aisg-aiml/lists"}